Integrated analysis of pediatric low-grade glioma: clinical implications and the path forward

See the article by Fukuoka et al in this issue, pp. 1474–1483.

Likely no aspect of clinical pediatric neuro-oncology has been so dramatically impacted both diagnostically and therapeutically by recent molecular discoveries as has that of pediatric low-grade gliomas (PLGGs) and closely associated neuronal and mixed neuronal glial tumors. In just over a decade, since the discovery by Pfister and colleagues that KIAA1549:BRAF oncogenic gene fusions underlie the majority of pilocytic astrocytomas (PAs), subsequently confirmed by multiple other laboratories across the world, knowledge concerning PLGG molecular pathogenesis has rapidly expanded.¹ Large, predominantly retrospective studies have disclosed that virtually all PAs and the vast majority of other PLGGs are driven by genetic alterations resulting in aberrant signaling of the Ras/mitogen-activated protein kinase (MAPK) pathway. Even more importantly, prospective clinical trials have shown that Ras/MAPK inhibitors, including drugs targeting mitogen-activated protein kinase kinase (MEK) and BRAF p.V600E inhibitors, can be used effectively to halt PLGG progression. As outlined in a series of consensus statements over only a 10-year period, management of PLGGs has dramatically shifted from considerations of whether it is worth the risk to biopsy infiltrative chiasmatic/hypothalamic lesions to general agreement that tumor molecular analysis is required for meaningful diagnosis in the majority of tumors that cannot be totally resected and that such information can be used to guide therapy.^2–4 Phase III studies are already under way in children with and without neurofibromatosis 1 (NF1)-associated LGGs comparing the efficacy of MEK inhibitors to standard chemotherapy, as well as other studies evaluating BRAF inhibitors, at times coupled with MEK inhibitors, for tumors harboring BRAF p.V600E mutations.^5,6

As the field has evolved, robust classification algorithms utilizing DNA-methylation arrays have been developed and retrospective studies in relatively large numbers of tumor-tissue samples have demonstrated that genome-wide epigenetic patterns can be used to subclassify PLGGs and neuroglial tumors.⁷ Such epigenetic classifiers have applied cluster analysis to group and essentially reconceptualize the 13 different types of World Health Organization (WHO)-recognized neuronal and mixed neuronal glial tumors. However, it should be noted that current WHO classification does not use PLGG molecular findings for tumor subtype identification.⁸

In this landscape, the paper by Fukuoka et al, published in this issue of Neuro-Oncology, puts into perspective the value of PLGG epigenetic analysis by DNA methylation.⁹ Work performed primarily from the laboratories at the Hospital for Sick Children in 152 PLGGs (supplemented by 70 cases from St Jude’s to assess the confidence of the Heidelberg classifier) were analyzed by combined genetic, morphological, and epigenetic assessment. DNA methylation separated PLGGs into 3 major clusters, and as others have shown, these clusters were greatly impacted by tumor location and had prognostic significance. A major finding was that the methylation classifier alone failed to reach a high confidence level across subgroups. Moreover, nearly 11% of pathologically diagnosed PLGGs with molecular alterations consistent with PLGGs were considered nontumor tissue by the classifier. The authors postulate that the multiple normal and reactive cells, which may be present in infiltrating pediatric low-grade biopsies, may be one reason for the classifier “failure.”

By comparison, in an expanded cohort of 27 pleomorphic xanthoastrocytoma–like tumors, as defined by the Heidelberg classifier, some of which had been histologically diagnosed as other tumors such as gangliogliomas and desmoplastic infantile gangliogliomas, there was a high degree of confidence. All but 3 tumors had both BRAF p.V600E mutation and cyclin-dependent kinase inhibitor 2A (CDKN2A) homozygous deletion, and pathologic grade, not molecular alteration, correlated with survival. Another novel finding of the Fukuoka et al study, of potential therapeutic importance, was that the degree of leukocyte/lymphocytic infiltration as suggested by DNA methylation, eloquently named the LUMP score, inversely correlated with prognosis in cluster 2B tumors (PLGGs enriched with BRAF p.V600E and CDKN2A homozygous deletions), as a high LUMP score was associated with a poorer prognosis.

Taken in total, this detailed analysis demonstrates both the power and potential limitations of DNA-methylation analysis of PLGGs if used in isolation and suggests that an integrated approach utilizing epigenetic assessment, genetic analysis, and histopathological examination results in optimal stratification. A caveat is that the pathological analysis was performed by very experienced neuropathologists and the conclusions drawn concerning the value of histopathological assessment may not be generalizable. An area that deserves future study is the tantalizing association between a relative increase in the lymphocytic infiltration and poorer outcome in BRAF p.V600E mutant LGGs, raising the possible utility of immunotherapy (such as immune check point inhibitors) for patients harboring LGGs with BRAF p.V600E and CDKN2A homozygous deletion; a PLGG subtype with a less favorable prognosis.³

Despite some of the limitations of methylation-based classifications when utilized in isolation, there is no question that methylation analysis has dramatically improved our understanding of pediatric LGGs and low-grade mixed neuronal and neuronal glial tumors; for many tumor types it remains likely the best means for specific diagnosis. Similar analysis may also, in time, lead to the identification of the cell or origin and cellular microenvironment of PLGGs and resultant novel therapies. An integrated approach is likely to identify more complexities and is also likely to be key to further unlocking the pathobiology of PLGGs and mixed neuronal glial tumors. As therapeutic approaches rapidly move forward, such more-in-depth understanding will be needed, as will be a better understanding of the impact of therapy and other factors critical in PLGG biology, such as senescence.¹⁰ How any form of new treatment impacts the long-term course of these often chronic diseases needs to be viewed in the perspective of not only 5-year progression-free survival, but of impact over decades of life. It is increasingly clear that targeting the Ras-MAPK underpinnings of PLGGs and neuroglial tumors is a remarkable advance and the likely basis of therapeutic interventions for years to come. How to best build on it remains an unanswered question.

Conflict of interest statement. Dr Packer is on the Advisory Committee for Novartis and AstraZeneca. There are no conflicts for Dr MacDonald.